CN111862079B - High-grade serous ovarian cancer recurrence risk prediction system based on image histology - Google Patents
High-grade serous ovarian cancer recurrence risk prediction system based on image histology Download PDFInfo
- Publication number
- CN111862079B CN111862079B CN202010755346.8A CN202010755346A CN111862079B CN 111862079 B CN111862079 B CN 111862079B CN 202010755346 A CN202010755346 A CN 202010755346A CN 111862079 B CN111862079 B CN 111862079B
- Authority
- CN
- China
- Prior art keywords
- image
- recurrence
- module
- histology
- tumor segmentation
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/12—Edge-based segmentation
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/50—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for simulation or modelling of medical disorders
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10004—Still image; Photographic image
- G06T2207/10012—Stereo images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30096—Tumor; Lesion
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Public Health (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Biomedical Technology (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Pathology (AREA)
- Primary Health Care (AREA)
- Epidemiology (AREA)
- Databases & Information Systems (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Radiology & Medical Imaging (AREA)
- Quality & Reliability (AREA)
- Magnetic Resonance Imaging Apparatus (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses an image histology-based high-level serous ovarian cancer recurrence risk prediction system which comprises T1 weighted enhanced image histology processing, T2 weighted image histology processing and information fusion. Wherein, the image histology processing mainly includes: three-dimensional tumor segmentation, image standardization, image histology feature extraction, feature normalization, feature screening, SMOTE resampling and classifier training; the information fusion is mainly used for fusing recurrence risk prediction probabilities output by T1 and T2 image histology processing, so that the accuracy of risk prediction is further improved.
Description
Technical Field
The invention relates to the technical field of computer-aided diagnosis, in particular to a high-level serous ovarian cancer recurrence risk prediction system based on an image histology method.
Background
High grade serous ovarian cancer (High-grade serous ovarian cancer, HGSOC) is the most common subtype of ovarian cancer, accounting for about 70%, with the vast majority of patients already in progress at the visit. Currently, the preferred treatment regimen is still supplemented with initial tumor cell debulking followed by post-operative platinum-based chemotherapy. Although the initial treatment efficiency can reach 80%, about 85% of patients can relapse tumor until drug resistance occurs, and the overall 5-year survival rate is only about 30%. Clinically, there is still a lack of effective and reliable markers to determine the risk of recurrence of tumors, a challenge for oncologists. Recent studies have shown that maintenance therapy based on PARPI inhibitors + bevacizumab can significantly extend Progression-free survival (PFS) of patients. Thus, early detection of HGSOC patients at high risk of recurrence, and first-line maintenance therapy, may bring potentially significant clinical benefit. Magnetic resonance imaging (Magnetic resonance imaging, MRI) has the advantages of high soft tissue contrast, multiple sequence and multiple parameter imaging, playing an important role in the assessment of ovarian cancer. Earlier studies have shown that morphological features based on conventional sequences and quantitative parameters based on functional sequences are of limited value in the prognosis prediction of ovarian cancer. In recent years, the application of image histology in the tumor field shows a good prospect and can effectively guide clinical decision making. Therefore, the model for predicting the recurrence risk of the HGSOC in the development period is constructed by extracting the MRI histology characteristics of the HGSOC primary focus and fusing clinical factors, and the model has important theoretical significance and application value.
Disclosure of Invention
The invention designs a high-level serous ovarian cancer recurrence risk prediction system based on a magnetic resonance image histology method by utilizing a T1 weighted enhanced image and a T2 weighted image, and realizes early and accurate prediction of recurrence risk of a high-level serous ovarian cancer patient.
The invention solves the technical problems by the following technical proposal:
the invention provides an image histology-based high-level serous ovarian cancer recurrence risk prediction system which is characterized by comprising a tumor segmentation module, an image standardization module, a feature extraction module, a feature normalization module, a feature screening module, a resampling module, a training module and an information fusion module;
the tumor segmentation module is used for respectively carrying out three-dimensional tumor segmentation on each T1 weighted enhanced image in the obtained T1 original sample and a primary tumor region in each T2 weighted image in the T2 original sample so as to obtain a T1 three-dimensional tumor segmentation image and a T2 three-dimensional tumor segmentation image;
the image normalization module is used for normalizing gray values of the T1 three-dimensional tumor segmentation image and the T2 three-dimensional tumor segmentation image to an optimal display range and normalization respectively, and resampling the T1 three-dimensional tumor segmentation image and the T2 three-dimensional tumor segmentation image respectively to normalize image resolution;
the feature extraction module is used for quantitatively calculating 1046 image histology features of each normalized T1 three-dimensional tumor segmentation image and T2 three-dimensional tumor segmentation image by using an image histology feature extraction package;
the feature normalization module is used for respectively carrying out normalization processing on each image group feature in each T1 three-dimensional tumor segmentation image and each T2 three-dimensional tumor segmentation image;
the feature screening module is used for screening the image histology features in each normalized T1 three-dimensional tumor segmentation image and each normalized T2 three-dimensional tumor segmentation image respectively to obtain a T1 image histology optimal feature set and a T2 image histology optimal feature set;
the resampling module is used for setting resampling probability according to the ratio of the recurrence and non-recurrence samples corresponding to the T1 original sample and the ratio of the recurrence and non-recurrence samples corresponding to the T2 original sample, and resampling the non-recurrence samples of the T1 image group optimal feature set and the non-recurrence samples of the T2 image group optimal feature set respectively, so that the number of the recurrence and non-recurrence samples of the T1 image group optimal feature set is basically consistent with the number of the recurrence and non-recurrence samples of the T2 image group optimal feature set;
the training module is used for selecting a classifier, and respectively constructing and training a corresponding prediction model by utilizing each resampled T1 image histology optimal feature set and T2 image histology optimal feature set, so as to respectively output recurrence risk probability P corresponding to each original sample T1 And P T2 ；
The information fusion module is used for fusing the recurrence risk probability P corresponding to each original sample T1 And P T2 And information fusion is carried out to screen out the optimal fusion model.
Preferably, the tumor segmentation module is used for performing three-dimensional tumor segmentation on the primary tumor region in the T1 weighted enhancement image and the T2 weighted image by utilizing ITK-Snap software and through a full-automatic or interactive segmentation mode.
Preferably, the image normalization module is configured to normalize gray values of the T1 three-dimensional tumor segmentation image and the T2 three-dimensional tumor segmentation image to an optimal display range according to a default window width and a default window level of each layer of image in the magnetic resonance sequence image by using a window width window level adjustment technology, unify the normalized gray values to [0,1200], and resample the T1 three-dimensional tumor segmentation image and the T2 three-dimensional tumor segmentation image by using a cubic spline difference algorithm to normalize image resolutions to [1mm, 5mm ] and [1mm, 8mm ] respectively.
Preferably, the feature extraction module is configured to use an image histology feature extraction package PyRadiomics to quantitatively calculate 1046 image histology features of each normalized T1 three-dimensional tumor segmentation image and T2 three-dimensional tumor segmentation image, and mainly includes 100 original image features, 258 LoG image features and 688 wavelet image features, and mainly covers 3D shape features, gray histogram features and texture features;
wherein, the texture features mainly comprise: gray level co-occurrence matrix texture features, gray level size region matrix texture features, gray level run length matrix texture features, adjacent gray level tone difference matrix texture features, and gray level dependent matrix texture features.
Preferably, the feature normalization module is used for performing normalization processing on each image group chemical feature in each T1 three-dimensional tumor segmentation image and each image group chemical feature in each T2 three-dimensional tumor segmentation image by using a min-max normalization method.
Preferably, the feature screening module is configured to use an L1 regularization feature selection method to screen the image histology features in each normalized T1 three-dimensional tumor segmentation image and each normalized T2 three-dimensional tumor segmentation image, so as to obtain a T1 image histology optimal feature set and a T2 image histology optimal feature set.
Preferably, the resampling module is configured to set a resampling probability according to a ratio of the recurrent sample and the unrepeated sample corresponding to the T1 original sample, resample the unrepeated sample in the T1 image group optimal feature set by using an SMOTE resampling method so that the number of recurrent and unrepeated samples in the T1 image group optimal feature set is substantially identical, set a resampling probability according to a ratio of the recurrent and unrepeated sample corresponding to the T2 original sample, and resample the unrepeated sample in the T2 image group optimal feature set by using an SMOTE resampling method so that the number of recurrent and unrepeated samples in the T2 image group optimal feature set is substantially identical.
Preferably, the training module is configured to select an SVM classifier, and respectively construct and train a corresponding prediction model by using each resampled T1 image histology optimal feature set and T2 image histology optimal feature set, so as to respectively output a recurrence risk probability P corresponding to each original sample T1 And P T2 。
Preferably, the information fusion module is configured to compare the recurrence risk probability P corresponding to each original sample T1 And P T2 Information fusion is carried out to obtain a plurality of fusion models, and an optimal fusion model is screened out of the fusion models;
the formulation of the information fusion strategy is as follows:
wherein P is T1 And P T2 The output probabilities of the prediction model based on the T1 weighted enhanced video and the prediction model based on the T2 weighted video are respectively represented, and min and max represent the minimum value and the maximum value of the two.
On the basis of conforming to the common knowledge in the field, the above preferred conditions can be arbitrarily combined to obtain the preferred examples of the invention.
The invention has the positive progress effects that:
the HGSOC recurrence risk prediction system based on MRI image histology mainly extracts the image characteristics in tumor at high flux, and builds a prediction model by using a machine learning classifier to realize early prediction of recurrence risk. The system can explore the association between the tumor imaging phenotype and the recurrence risk of the patient on the basis of mining the heterogeneity information in the tumor, realize the early monitoring of the recurrence risk of HGSOC, and assist the clinic in decision making.
Compared with a recurrence risk prediction model of high-level serous ovarian cancer reported in current domestic and foreign documents, the prediction model based on magnetic resonance image histology feature analysis is provided, the image information of a T1 weighted enhancement image and a T2 weighted image can be fused, the recurrence risk of the high-level serous ovarian cancer is predicted noninvasively, and the accuracy of recurrence risk prediction is improved.
Drawings
FIG. 1 is a block diagram of a high-level serous ovarian cancer recurrence risk prediction system based on image histology according to a preferred embodiment of the present invention.
FIG. 2 is a flowchart of an embodiment of an image histology processing module.
FIG. 3 is a diagram showing a pre-experimental result of a preferred embodiment of the present invention.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is apparent that the described embodiments are some embodiments of the present invention, but not all embodiments of the present invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
As shown in fig. 1, the embodiment provides a high-level serous ovarian cancer recurrence risk prediction system based on image histology, which comprises a tumor segmentation module 1, an image standardization module 2, a feature extraction module 3, a feature normalization module 4, a feature screening module 5, a resampling module 6, a training module 7 and an information fusion module 8.
The tumor segmentation module 1 is used for carrying out three-dimensional tumor segmentation on primary tumor areas in each T1 weighted enhanced image in the obtained T1 original sample and each T2 weighted image in the T2 original sample by utilizing ITK-Snap software (http:// www.itksnap.org /) and through a full-automatic or interactive segmentation mode, and drawing boundaries of ovarian tumors so as to obtain a T1 three-dimensional tumor segmentation image and a T2 three-dimensional tumor segmentation image.
The image normalization module 2 is configured to normalize gray values of the T1 three-dimensional tumor segmentation image and the T2 three-dimensional tumor segmentation image to an optimal display range according to a default window width and a default window level of each layer of image in the magnetic resonance sequence image by using a window width window level adjustment technology, unify the normalized gray values to [0,1200], and resample the T1 three-dimensional tumor segmentation image and the T2 three-dimensional tumor segmentation image by using a cubic spline difference algorithm to normalize image resolutions to [1mm, 5mm ] and [1mm, 8mm ] respectively.
The feature extraction module 3 is configured to use an image histology feature extraction package pyradius (https:// tissues, readthes. Io /) to quantitatively calculate 1046 image histology features of each normalized T1 three-dimensional tumor segmentation image and T2 three-dimensional tumor segmentation image, and mainly includes 100 original image features, 258 LoG image features and 688 wavelet image features, and mainly covers 3D shape features, gray histogram features and texture features.
Wherein, the texture features mainly comprise: gray level co-occurrence matrix texture features, gray level size region matrix texture features, gray level run length matrix texture features, adjacent gray level tone difference matrix texture features, and gray level dependent matrix texture features.
The feature normalization module 4 is used for performing normalization processing on each image group feature in each T1 three-dimensional tumor segmentation image and each image group feature in each T2 three-dimensional tumor segmentation image by using a min-max normalization method.
The feature normalization mainly uses a min-max normalization (also called as dispersion numerical normalization) method to normalize different types of image features to be within a range of 0-1 so as to improve the convergence speed of the model and the robustness and the accuracy of the training model. The normalized calculation formula for a certain image feature X is as follows:
wherein X is normalization And (3) representing the normalized characteristic value, wherein X is the original characteristic value, max is the maximum value of the original characteristic, and min is the minimum value of the original characteristic.
The feature screening module 5 is configured to use an L1 regularization feature selection method to screen the image histology features in each normalized T1 three-dimensional tumor segmentation image and each normalized T2 three-dimensional tumor segmentation image, so as to obtain a T1 image histology optimal feature set and a T2 image histology optimal feature set.
The feature screening is mainly to screen image features with better classification capability from thousands of image histology features by using an L1 regularization (Lasso regression) feature selection method so as to eliminate noise features and associated features, reduce training expenditure, improve the accuracy of classification model training and reduce overfitting.
The resampling module 6 is configured to set a resampling probability according to a ratio of the recurrent and unrepeated samples corresponding to the T1 original sample, resample the unrepeated samples in the T1 image group optimal feature set by using an SMOTE resampling method so that the number of recurrent and unrepeated samples in the T1 image group optimal feature set is substantially identical, set a resampling probability according to a ratio of the recurrent and unrepeated samples corresponding to the T2 original sample, and resample the unrepeated samples in the T2 image group optimal feature set by using an SMOTE resampling method so that the number of recurrent and unrepeated samples in the T2 image group optimal feature set is substantially identical.
The SMOTE resampling is mainly used for balancing the balance of two groups of data in a training sample, and is used for oversampling the characteristic values of the categories (non-recurrent patients) with fewer samples in the training set, and synthesizing new characteristic samples to relieve the unbalance of the two types of samples. The specific algorithm flow is as follows: (1) eigenvalues F for each non-recurring sample Non-Recurrence And calculating the distances from the Euclidean distance to all samples in the minority sample set by taking the Euclidean distance as a standard to obtain the K nearest neighbor. (2) Setting a sampling ratio according to the sample imbalance ratio to determine the sampling rate N, for each unrepeated sample F Non-Recurrence Randomly selecting from its K-neighborsSeveral samples, assuming selected neighbors as F n . (3) For each randomly selected neighbor F n Constructing a new sample as F according to the following formula with the original sample new ＝F Non-Recurrence +rand(0,1)×|F Non-Recurrence -F n |。
The training module 7 is configured to select an SVM classifier, and respectively construct and train a corresponding prediction model by using each resampled T1 image histology optimal feature set and T2 image histology optimal feature set, so as to respectively output a recurrence risk probability P corresponding to each original sample T1 And P T2 。
Classifier training is mainly to train a support vector machine (Support Vector Machine, SVM) classifier to construct a classification model by using the resampled sample characteristics, and predict the recurrence probability of high-level serous ovarian cancer. And under the condition of small sample size, testing the performance of the classification model by adopting a leave-one-out cross-validation method.
The information fusion module 8 is configured to compare the recurrence risk probability P corresponding to each original sample T1 And P T2 And carrying out information fusion to obtain a plurality of fusion models, and screening out the optimal fusion model from the fusion models.
The formulation of the information fusion strategy is as follows:
wherein P is T1 And P T2 The output probabilities of the prediction model based on the T1 weighted enhanced video and the prediction model based on the T2 weighted video are respectively represented, and min and max represent the minimum value and the maximum value of the two.
The effect of the invention is further illustrated by the following experiments:
1. experimental conditions:
the experimental data were from 117 patients in the accessory oncology hospital at the university of double denier. Of these, 90 patients had relapsed after surgery and 27 patients had not relapsed by follow-up for at least 18 months. Under the python 3.7 environment, an open source library such as scikit-learn, pyradiomics, scipy is used for constructing a high-level serous ovarian cancer recurrence risk prediction model based on image histology.
2. Experimental results and results analysis
Referring to FIG. 3, the AUC values of the model constructed using T1 weighted enhancement images and T2 weighted images were 0.79+ -0.04 (95% CI: [0.69,0.86 ]) and 0.74+ -0.05 (95% CI: [0.63,0.83 ]), respectively, with the AUC of the model fused with two different image characteristics reaching 0.85+ -0.04 (95% CI: [0.75,0.90 ]). Compared with a model using a single image feature, the fusion model can effectively improve the prediction accuracy (p < 0.05) of the recurrence risk of HGSOC in the development period. Compared with related similar researches in recent years, the novel method is superior to experimental results of similar research high-grade serous ovarian cancer recurrence risk prediction methods, so that the method has certain superiority.
While specific embodiments of the invention have been described above, it will be appreciated by those skilled in the art that these are by way of example only, and the scope of the invention is defined by the appended claims. Various changes and modifications to these embodiments may be made by those skilled in the art without departing from the principles and spirit of the invention, but such changes and modifications fall within the scope of the invention.
Claims (9)
1. The high-level serous ovarian cancer recurrence risk prediction system based on image histology is characterized by comprising a tumor segmentation module, an image standardization module, a feature extraction module, a feature normalization module, a feature screening module, a resampling module, a training module and an information fusion module;
the tumor segmentation module is used for respectively carrying out three-dimensional tumor segmentation on each T1 weighted enhanced image in the obtained T1 original sample and a primary tumor region in each T2 weighted image in the T2 original sample so as to obtain a T1 three-dimensional tumor segmentation image and a T2 three-dimensional tumor segmentation image;
the image normalization module is used for normalizing gray values of the T1 three-dimensional tumor segmentation image and the T2 three-dimensional tumor segmentation image to an optimal display range and normalization respectively, and resampling the T1 three-dimensional tumor segmentation image and the T2 three-dimensional tumor segmentation image respectively to normalize image resolution;
the feature extraction module is used for quantitatively calculating 1046 image histology features of each normalized T1 three-dimensional tumor segmentation image and T2 three-dimensional tumor segmentation image by using an image histology feature extraction package;
the feature normalization module is used for respectively carrying out normalization processing on each image group feature in each T1 three-dimensional tumor segmentation image and each T2 three-dimensional tumor segmentation image;
the feature screening module is used for screening the image histology features in each normalized T1 three-dimensional tumor segmentation image and each normalized T2 three-dimensional tumor segmentation image respectively to obtain a T1 image histology optimal feature set and a T2 image histology optimal feature set;
the resampling module is used for setting resampling probability according to the ratio of the recurrence and non-recurrence samples corresponding to the T1 original sample and the ratio of the recurrence and non-recurrence samples corresponding to the T2 original sample, and resampling the non-recurrence samples of the T1 image group optimal feature set and the non-recurrence samples of the T2 image group optimal feature set respectively, so that the number of the recurrence and non-recurrence samples of the T1 image group optimal feature set is basically consistent with the number of the recurrence and non-recurrence samples of the T2 image group optimal feature set;
the training module is used for selecting a classifier, and respectively constructing and training a corresponding prediction model by utilizing each resampled T1 image histology optimal feature set and T2 image histology optimal feature set, so as to respectively output recurrence risk probability P corresponding to each original sample T1 And P T2 ；
The information fusion module is used for fusing the recurrence risk probability P corresponding to each original sample T1 And P T2 And information fusion is carried out to screen out the optimal fusion model.
2. The imaging-histology-based high-level serous ovarian cancer recurrence risk prediction system according to claim 1, wherein the tumor segmentation module is configured to perform three-dimensional tumor segmentation on primary tumor regions in the T1-weighted enhanced image and the T2-weighted image by using ITK-Snap software and through a full-automatic or interactive segmentation method, respectively.
3. The imaging-based high-level serous ovarian cancer recurrence risk prediction system according to claim 1, wherein the image normalization module is configured to normalize gray values of the T1 three-dimensional tumor segmentation image and the T2 three-dimensional tumor segmentation image to an optimal display range according to a default window width and a default window level of each layer of images in the magnetic resonance sequence images, respectively, and then unify the normalized to [0,1200], and resample the T1 three-dimensional tumor segmentation image and the T2 three-dimensional tumor segmentation image to normalize image resolutions to [1mm, 5mm ] and [1mm, 8mm ], respectively, using a cubic spline difference algorithm.
4. The imaging-based high-level serous ovarian cancer recurrence risk prediction system according to claim 1, wherein the feature extraction module is configured to use imaging-based feature extraction package PyRadiomics to quantitatively calculate 1046 imaging-based features of each normalized T1 and T2 three-dimensional tumor segmentation image, mainly including 100 original image features, 258 LoG image features and 688 wavelet image features, and mainly covering 3D shape features, gray histogram features and texture features;
wherein, the texture features mainly comprise: gray level co-occurrence matrix texture features, gray level size region matrix texture features, gray level run length matrix texture features, adjacent gray level tone difference matrix texture features, and gray level dependent matrix texture features.
5. The imaging-histology-based high-level serous ovarian cancer recurrence risk prediction system according to claim 1, wherein the feature normalization module is configured to normalize each of the imaging histology features in each of the T1 three-dimensional tumor segmentation image and the T2 three-dimensional tumor segmentation image by using a min-max normalization method.
6. The imaging-based high-level serous ovarian cancer recurrence risk prediction system according to claim 1, wherein the feature screening module is configured to screen the image histology features in each normalized T1 three-dimensional tumor segmentation image and T2 three-dimensional tumor segmentation image by using an L1 regularization feature selection method, so as to obtain a T1 image histology optimal feature set and a T2 image histology optimal feature set.
7. The high-grade serous ovarian cancer recurrence risk prediction system according to claim 1, wherein the resampling module is configured to set a resampling probability according to a ratio of recurrence and non-recurrence samples corresponding to the T1 original sample, resample the non-recurrence samples in the T1 image set of optimal features using the SMOTE resampling method so that the number of recurrence and non-recurrence samples in the T1 image set of optimal features is substantially identical, set a resampling probability according to a ratio of recurrence and non-recurrence samples corresponding to the T2 original sample, and resample the non-recurrence samples in the T2 image set of optimal features using the SMOTE resampling method so that the number of recurrence and non-recurrence samples in the T2 image set of optimal features is substantially identical.
8. The image-based high-level serous ovarian cancer recurrence risk prediction system as set forth in claim 1, wherein the training module is configured to select an SVM classifier, and respectively construct and train a corresponding prediction model using each of the resampled T1 image group optimal feature set and the resampled T2 image group optimal feature set, thereby respectively outputting recurrence risk probability P corresponding to each of the original samples T1 And P T2 。
9. The imaging-based high-grade serous ovarian cancer recurrence risk prediction system as claimed in claim 1, wherein the information fusion module is configured to map each raw sample with a recurrence risk probability P T1 And P T2 Information fusion is carried out to obtain a plurality of fusion models, and the fusion models are selected from the fusion modelsScreening out an optimal fusion model;
the formulation of the information fusion strategy is as follows:
wherein P is T1 And P T2 The output probabilities of the prediction model based on the T1 weighted enhanced video and the prediction model based on the T2 weighted video are respectively represented, and min and max represent the minimum value and the maximum value of the two.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010755346.8A CN111862079B (en) | 2020-07-31 | 2020-07-31 | High-grade serous ovarian cancer recurrence risk prediction system based on image histology |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010755346.8A CN111862079B (en) | 2020-07-31 | 2020-07-31 | High-grade serous ovarian cancer recurrence risk prediction system based on image histology |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111862079A CN111862079A (en) | 2020-10-30 |
| CN111862079B true CN111862079B (en) | 2023-09-22 |
Family
ID=72945774
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010755346.8A Active CN111862079B (en) | 2020-07-31 | 2020-07-31 | High-grade serous ovarian cancer recurrence risk prediction system based on image histology |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111862079B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112561869B (en) * | 2020-12-09 | 2021-11-02 | 深圳大学 | Pancreatic neuroendocrine tumor postoperative recurrence risk prediction method |
| CN113488170B (en) * | 2021-07-02 | 2023-07-25 | 温州医科大学 | Method for constructing acute pre-uveitis recurrence risk prediction model and related equipment |
| CN113902724B (en) * | 2021-10-18 | 2022-07-01 | 广州医科大学附属肿瘤医院 | Method, device, equipment and storage medium for classifying tumor cell images |
| CN116230227A (en) * | 2023-02-27 | 2023-06-06 | 复旦大学附属肿瘤医院 | Lung cancer brain metastasis risk prediction method |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105316402A (en) * | 2015-04-02 | 2016-02-10 | 复旦大学附属肿瘤医院 | MRNA [messenger RNA (ribonucleic acid)] and lncRNA (long non-coding RNA) combination model for predicting prognosis and chemotherapy sensitivity of patients suffering from triple-negative breast cancer and application of mRNA and lncRNA combination model |
| CN109685767A (en) * | 2018-11-26 | 2019-04-26 | 西北工业大学 | A kind of bimodal brain tumor MRI dividing method based on Cluster-Fusion algorithm |
| CN110265095A (en) * | 2019-05-22 | 2019-09-20 | 首都医科大学附属北京佑安医院 | For HCC recurrence and construction method and the application of the prediction model and nomogram of RFS |
| CN111353998A (en) * | 2020-05-13 | 2020-06-30 | 温州医科大学附属第一医院 | Tumor diagnosis and treatment prediction model and device based on artificial intelligence |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10980519B2 (en) * | 2015-07-14 | 2021-04-20 | Duke University | Systems and methods for extracting prognostic image features |
| US10846367B2 (en) * | 2017-09-15 | 2020-11-24 | Case Western Reserve University University | Predicting recurrence in early stage non-small cell lung cancer (NSCLC) with integrated radiomic and pathomic features |
-
2020
- 2020-07-31 CN CN202010755346.8A patent/CN111862079B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105316402A (en) * | 2015-04-02 | 2016-02-10 | 复旦大学附属肿瘤医院 | MRNA [messenger RNA (ribonucleic acid)] and lncRNA (long non-coding RNA) combination model for predicting prognosis and chemotherapy sensitivity of patients suffering from triple-negative breast cancer and application of mRNA and lncRNA combination model |
| CN109685767A (en) * | 2018-11-26 | 2019-04-26 | 西北工业大学 | A kind of bimodal brain tumor MRI dividing method based on Cluster-Fusion algorithm |
| CN110265095A (en) * | 2019-05-22 | 2019-09-20 | 首都医科大学附属北京佑安医院 | For HCC recurrence and construction method and the application of the prediction model and nomogram of RFS |
| CN111353998A (en) * | 2020-05-13 | 2020-06-30 | 温州医科大学附属第一医院 | Tumor diagnosis and treatment prediction model and device based on artificial intelligence |
Non-Patent Citations (1)
| Title |
|---|
| 基于多参数磁共振成像影像组学特征的膀胱癌复发预测;杜鹏;王焕军;石宇强;刘洋;卢虹冰;郭燕;潘维滨;张曦;徐肖攀;;中国医学装备(11);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111862079A (en) | 2020-10-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111862079B (en) | High-grade serous ovarian cancer recurrence risk prediction system based on image histology | |
| Papanikolaou et al. | How to develop a meaningful radiomic signature for clinical use in oncologic patients | |
| US10235755B2 (en) | High-throughput adaptive sampling for whole-slide histopathology image analysis | |
| Tiwari et al. | Multi-kernel graph embedding for detection, Gleason grading of prostate cancer via MRI/MRS | |
| Xu et al. | Using transfer learning on whole slide images to predict tumor mutational burden in bladder cancer patients | |
| Xu et al. | Differentiating benign from malignant renal tumors using T2‐and diffusion‐weighted images: A comparison of deep learning and radiomics models versus assessment from radiologists | |
| US10956795B2 (en) | Predicting recurrence in early stage non-small cell lung cancer (NSCLC) using spatial arrangement of clusters of tumor infiltrating lymphocytes and cancer nuclei | |
| US20220108123A1 (en) | Tissue microenvironment analysis based on tiered classification and clustering analysis of digital pathology images | |
| Li et al. | [Retracted] Automated Diagnosis and Localization of Melanoma from Skin Histopathology Slides Using Deep Learning: A Multicenter Study | |
| Chaudhury et al. | Deep transfer learning for IDC breast cancer detection using fast AI technique and Sqeezenet architecture | |
| yahia Ibrahim et al. | An enhancement technique to diagnose colon and lung cancer by using double CLAHE and deep learning | |
| Tenali et al. | Oral Cancer Detection using Deep Learning Techniques | |
| Singh et al. | STRAMPN: Histopathological image dataset for ovarian cancer detection incorporating AI-based methods | |
| CN115206497A (en) | Lung adenocarcinoma subtype analysis method based on deep learning method and non-diagnosis purpose and interpretation method thereof | |
| Agarwal et al. | OCS-Net: a novel approach for oral carcinoma stage classification using generative adversarial networks and genetic algorithm-based convolutional neural networks | |
| EP4348596A1 (en) | Detecting tertiary lymphoid structures in digital pathology images | |
| CN113889235A (en) | Unsupervised feature extraction system for three-dimensional medical image | |
| Ghouri et al. | Radiomic features extraction based on genetic algorithm | |
| RS | CoC-ResNet-classification of colorectal cancer on histopathologic images using residual networks | |
| Liu et al. | Identification of lymph node metastasis in pre‐operation cervical cancer patients by weakly supervised deep learning from histopathological whole‐slide biopsy images | |
| Maimone et al. | Comparison of Machine and Deep Learning models for automatic segmentation of prostate cancers on multiparametric MRI | |
| Rashid et al. | Automatic detection of ovarian cancer based on improved DWT transformation | |
| Pirzad Mashak et al. | A new approach for data augmentation in a deep neural network to implement a monitoring system for detecting prostate cancer in MRI images 1 | |
| Karsaz | Diabetic retinopathy screening using improved support vector domain description: a clinical study | |
| Zhang | Deep Radiomics Analytics Pipeline for Prognosis of Pancreatic Ductal Adenocarcinoma |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |